Arm impairment is a common condition that contributes substantially to disability in the U.S. and around the world. For example, it is estimated that 80% of the 700,000 individuals who survive a stroke each year require arm therapy. Intensive rehabilitation reduces long-term arm impairment, especially when performed early after stroke. Unfortunately, time with a therapist is limited during inpatient rehabilitation and individuals with severe impairment lack motivating and effective tools to increase the dose of therapy they receive. To address this problem, we have developed LARA: an assistive, lever-drive wheelchair specifically designed to provide arm rehabilitation through repetitive elbow/shoulder flexion/extension exercise. In pilot studies, we found 12 out of 12 individuals with severe arm impairment were able to use LARA with their impaired arm to move 50 feet overground in their wheelchair and to play engaging therapy-based computer games while exercising in LARA's stationary mode. Thus, by using LARA as their wheelchair for the several hours per day they normally spend in a wheelchair, inpatients will be able to use their impaired arm to ambulate to their therapy appointments, the cafeteria, etc., rather than being passively pushed by a caregiver or by using only their good arm and foot, as is now the case. They will also be able to easily access the therapy-based games in their room or in the clinic common room. They will thus achieve thousands more rehabilitation practice movements than is current practice, improving their arm movement recovery. Our aims for Phase II are to: 1) Develop a production model of LARA suitable for commercialization (months 1-24). We will modify our functional prototype to be robust, maneuverable, easy to use, and efficiently manufactured in an FDA-compliant way, and will complete development of an intuitive, easy-to-use game interface that automatically quantifies and logs the user's movement and use habits. 2) Perform a randomized controlled trial of the device with sub-acute stroke patients (months 12-24). Individuals with subacute stroke are a large potential user group. Through a consortium with UC Irvine, we will randomize 44 individuals with a subacute stroke to use LARA or receive additional conventional therapy during their inpatient stay. We hypothesize that the individuals who use LARA to exercise and ambulate to their appointments will improve their Fugl-Meyer score by a clinically significant amount (7 points) more the control group as assessed at a three month follow-up. 3) Solicit feedback on usability of the device in normal clinical practice (months 12-24). We will distribute LARA to two leading rehabilitation units, and study how these units incorporate the device into their patient care. This Phase II project will result in a commercially ready, clinically-validated product that we will produce and sell.